APBRmetrics

Well, that is a discussion we had 2 years ago. Miller and Sanjurjo explained why the "hot hand" was not found in reallife experiments, because the statistical analysis applied was inappropiate. To be more specific: When selecting only shots after a certain sequence of made or missed shots,...

It's because games with higher roster turnover rates are found more in 2000s and 2010s and almost every metric's prediction success gets better when we get closer to our time. I just ran a ADF on that data and there is no such trend at all (talking about the bolded part). You just picked a sample w...

permaximum wrote:The third was true and very helpful. I thanked for it.

And that was info given by someone else before and I just repeated it ... I should have known where this was going ...

1. I will generate a "minutes from <250MP players / roster turnover chart" for you. I believe 250 MP cutoff should be more than sufficient and PM based metrics shouldn't have any problem with it because it stays roughly the same for any roster turnover rate so the line should be straight ...

If one accepts what permaximum presents, then RPM and other complex metrics that go beyond simple linear accounting might be less reliable when considering trades than simple linear accounting. Well, I would say the conclusion should be that such metrics have trouble with players for whom very litt...

First: Which RAPM version did you use? Prior informed RAPM does better in such tests than 1yr RAPM. Regarding your latest response: 1. I think you should check the data again, because what Nathan said is likely true. Also, when I tested that on season basis (using y-1 data to "predict" SRS...

I think you missed the point there, and I'm surprised you're defending PER at all. Did you notice I was ignoring FTs too? The point was showing how the metric responds to points and FG%. Does this:"Using Brook Lopez 2016 as an example, in which he takes 400 more FGA while converting 140 of tho...

http://fansided.com/2017/01/31/nylon-calculus-reinventing-per/ I saw this article today and saw a typical misleading argument in regard to the "break-even fg%" in PER. Using Brook Lopez 2016 as an example, in which he takes 400 more FGA while converting 140 of those additional shots, which...

> Marg=data$Margin > Poss=data$Poss > data$Margin=NULL > data$Poss=NULL > x=data.matrix(data) > lambda=cv.glmnet(x, Marg, weights=Poss, nfolds=10) > lambda.min=lambda$lambda.min > ridge=glmnet(x, Marg, family=c("gaussian"), Poss, alpha=0, standardize=FALSE, lambda=lambda.min) > coef(ridge...

Unfortunately as you know only glmnet provides weighted ridge regression in R. Indeed, I'm not aware of a package besides glmnet which would allow weighted ridge regression. But you could still write down the equations and then solve it (the matrix package contains a function solve()), and you can ...

It's been a quite some time but I recall I "probably" used standardize=FALSE anyways. I can't remember it correctly, but I think you posted some numbers and there were a couple of low possessions players on top of the list. I got similar results with glmnet and couldn't really figure out ...

I believe in one of my debates with mystic he acknowledged this issue with glmnet as well. Yes, I had similar issues, but it turned out that the documentation just isn't really good. glmnet has standardize set to TRUE by default. When using glmnet for RAPM calculations someone should use standardiz...

Mike G wrote:Is that logically true? Does a 3-game comeback last year make it less likely this year?

Obviously not. That sounds like a typical gambler's fallacy.

Also, the Warriors started the series with having HCA. Not many teams which were down 1-3 and lost had that luxury.

I conveyed a thought, Mike G, as always gave me his in-depth insight and seemed to understand me fine without me trying to impress him, or Crow, or J.E., or DSmok1 or anyone else with my knowledge of win expectancy/probability jargon. Win probability for the Warriors at this point: 0.178 Before the...

1andDunn wrote: Has anyone ever accounted for this (perhaps by putting substitutions as the last events) or is the overall impact marginal?

As far as I know everyone is accounting for this.